The Evidence on Teaching

Effects of a research-based intervention to improve seventh-grade students' proportional problem solving: A cluster randomized trial

Jitendra AK, Harwell MR, Dupuis DN, Karl SR, Lein AE, Simonson G, Slater SC · 2015

grade Brctdeveloper-ledfailed
Sample
1,999 students / 82 teachers and classrooms / 50 school districts
Population
US 7th graders across 50 districts.
Design
The scale-up of the 2013 efficacy trial, with cluster-correct multilevel analysis and observed implementation fidelity. Crucially, this trial DID administer an independent standardized outcome — the Process and Applications subtest of the Group Mathematics Assessment and Diagnostic Evaluation (GMADE).
Key findings
The honest at-scale magnitude for schema-based instruction is ~0.25-0.35 on trained content, not the g > 1.2 of the efficacy trial. And the one independent standardized measure ever administered in this series showed nothing, which is the series' central unresolved weakness.
Genetic confound
Cluster-randomized across 50 districts; selection is not the issue.
Replication notes
Same lab, same programme, four times the classrooms: g = 1.24 becomes ES ~0.32. Corrects an earlier claim in this archive that no independent standardized outcome was ever administered in this series — one was, and it was null.

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Proportional problem solving (researcher-built measure)ES~0.32 at posttest; ~0.25 at the 9-week retention testresearcher-designedend of unit and +9 weeksbusiness-as-usualunder-1yrdomain-skill
GMADE Process and Applications subtest (independent standardized measure)ESno treatment effectstandardizedend of unitbusiness-as-usualend-of-treatmentdomain-skill

Cited by